[freespace2] a problem that has me stumped!

[tigital]

>hiya,
>
>...after jumping the gun and releasing an OS X diff, I'm trying to 
>move ahead with the port and get the byteswapping to work with model 
>loading...fortunately, most of the that is done, but I'm now stuck 
>on what to do with a coupla struct data fields...
>
>the polymodel struct has an element "*submodel" of type bsp_info, 
>which is another struct...this bsp_info has a ubyte *bsp_data, which 
>is specifically passed to many model functions, which then use a 
>series of dereferencing define's to access certain fields:
>
>#define w(p) (*((int *) (p)))
>#define vp(p) ((vector *)(p))
>#define fl(p) (*((float *)(p)))
>
>
>...unfortunately, the fields are offsets from the beginning of the 
>"ubyte *bsp_data", so, as an example, with Textured poly's the 
>following offsets apply:
>
>+0	int	id
>+4	int	size
>+8	vector	normal
>+20	vector	center
>+32	float	radius
>+36	int	nverts
>+40	int	tmap_num
>+44	nverts*(model_tmap_vert)	vertlist (n,u,v)
>
>...so, the code uses vp(p+8) and vp(p+20) to access the vector 
>fields...so far I've been able to byteswap the w(p) and fl(p) 
>functions at the point of their calling(ex. chunk_size = 
>SDL_SwapLE32( w(p+4));), but I'm unsure how to access the vector 
>elements...a vector is as follows:
>
>typedef struct vector {
>	union {
>		struct{
>			float x,y,z;
>		}xyz;
>		float a1d[3];
>	};
>} vector;
>
>...anyone think they can shed some insight?
>
>tanx,
>
>jamie

Me again,

...the following compiles, but I don't know if trying to replace all 
44 occurrences with code like this is the best way to do it; seems 
like it should be done once, like at the model_read(), but there are 
four different types of bsp_data:  Point list, Flat Poly, Textured 
Poly, and Sortnorms (these are found in modelinterp.cpp, 
modeloctant.cpp, and modelcollide.cpp, just before the functions they 
are used in)...

             vector *tmp = vp(p+8);
             vector *tmp2 = vp(p+20);
             tmp->xyz.x = LoadLEFloat( &tmp->xyz.x );
             tmp->xyz.y = LoadLEFloat( &tmp->xyz.y );
             tmp->xyz.z = LoadLEFloat( &tmp->xyz.z );
             tmp2->xyz.x = LoadLEFloat( &tmp2->xyz.x );
             tmp2->xyz.y = LoadLEFloat( &tmp2->xyz.y );
             tmp2->xyz.z = LoadLEFloat( &tmp2->xyz.z );

...Seems like this could really slow down model rendering if this 
were done 44 times for each frame, but I don't know if that the 
functions that use vp(p) are used only in model loading, or are 
constantly called...

Help!

jamie